A credit union uses a generative model to draft explanations of loan eligibility for its underwriters. The compliance team wants the design to reflect responsible AI practices. Which choice best does that?
Show answer and explanation
Answer: D
High-impact decisions such as loan eligibility need human accountability, so the model should advise rather than decide. An audit trail of prompts and outputs supports review and investigation of disputes, and a sign-off step keeps a named underwriter responsible for the final outcome. An appeal path lets applicants correct errors. Changing creativity settings or hiding AI involvement does not address governance at all.
Why the other options are wrong
- A. Automatic approvals remove human accountability from a high-impact decision. Applicants cannot be assured of fairness, and the institution cannot correct errors that the model introduces.
- B. Hiding AI involvement undermines transparency, a core responsible AI principle. Users should know when automated tooling shaped a decision that affects them.
- C. Lower variability makes outputs more repeatable but does not guarantee correctness or fairness. Consistently wrong answers are still wrong, and this change provides no oversight.